Re: Syntax error in type_traits

[Jesper Dybdal <jd-debian-user@dybdal.dk>]

On 2025-01-16 13:07, Greg Wooledge wrote:

> On Thu, Jan 16, 2025 at 12:38:29 +0100, Jesper Dybdal wrote:
> > There is a syntax error (an erroneous ">" character) in
> >     /usr/include/c++/12/type_traits
> > in Bookworm.   So compiling any program that includes the <string> header
> > fails:
> >
> > echo "#include <string>" >error.cpp
> > g++ -c error.cpp
> >
> > Results in:
> >
> > > In file included from /usr/include/c++/12/bits/char_traits.h:42,
> > >                  from /usr/include/c++/12/string:40,
> > >                  from error.cpp:1:
> > > /usr/include/c++/12/type_traits:2614:11: error: expected identifier
> > > before ‘>’ token
> > >  2614 |     using >enable_if_t = typename enable_if<_Cond, _Tp>::type;
> > >       |           ^
> > >         
> >       
> This is not the case on my bookworm system (Debian 12.9, libstdc++-12-dev
> package version 12.2.0-14).
>
> hobbit:~$ sed -n 2614p /usr/include/c++/12/type_traits
>     using enable_if_t = typename enable_if<_Cond, _Tp>::type;
>
> What version of the libstdc++-12-dev package do you have?
>     
    jd@nuser:~$ apt list libstdc++-12-dev
    libstdc++-12-dev/stable,now 12.2.0-14 amd64 [installed,automatic]


> Does reinstalling that package fix the file?
>     
    Yes, it does!
    !
    
> Honestly, it's hard to imagine that this bug wouldn't have been caught,
> if it were widespread.  I have to think that hundreds of people compile
> C++ programs with <string> on Debian every day.
>     
    I was quite surprised too - glad  to hear that others do not have
    that problem.
> This feels more like a local file modification to me (either by human
> error, or file corruption, though the latter seems unlikely as well,
> since this is an *added* character, not a modified character).
>     
    Exactly
    
    I could have sworn that I verified yesterday that that header file
    had modification and change times that seemed plausible together
    with those of other files there.  But this morning (before the
    reinstall) it has a modification time of yesterday at 21:50.  And
    the directory has that same time also, but no other files there have
    been modified.
    
    But how I can have managed to add a byte in that file beats me.  I
    didn't even know that there was a file with that name, and I had not
    been near the directory before I saw the error message. And I had
    certainly not had it open in an editor.
    
    And about an hour ago, when I checked the file times, the other
    files in that directory had different, but old times (several with
    the same time, and several with other "same times"),  But now, after
    the reinstall, they all have 2023-01-08 as modification time.
    
    There is obviously nothing wrong with the package.  Sorry for
    bothering you all.
    
    But I am deeply confused about what is happening to my system.  
    
    Thanks to you and Andy,
    Jesper
-- 
Jesper Dybdal
https://www.dybdal.dk